Sorry, you need to enable JavaScript to visit this website.

Random forests analysis: A useful tool for defining the relative importance of environmental conditions on crown defoliation

TitoloRandom forests analysis: A useful tool for defining the relative importance of environmental conditions on crown defoliation
Tipo di pubblicazioneArticolo su Rivista peer-reviewed
Anno di Pubblicazione2014
AutoriVitale, M., Proietti C., Cionni Irene, Fischer R., and De Marco Alessandra
RivistaWater, Air, and Soil Pollution
Parole chiaveair monitoring, Air pollution, Air pollution effects, air pollution indicator, Ammonia, article, Assessment and monitoring, atmospheric deposition, atmospheric pollution, canopy architecture, classification, Classification (of information), classifier, Climate change, controlled study, Critical load, Decision trees, defoliation, Environmental conditions, Environmental monitoring, Fagus sylvatica, forest management, forestry, Information Retrieval, Meteorology, N deposition, Nitrogen, nitrogen oxide, nonhuman, Norway spruce, oak, Picea abies, Pinus sylvestris, plant age, pollution effect, Quercus ilex, Quercus petraea, random forest, Random forests, scots pine, Statistical classifier, tree crown

Defoliation is one of the most important parameters monitored in the International Cooperative Programme on Assessment and Monitoring of Air Pollution Effects on Forests (ICP Forests). Defoliation is an indicator for forest health and vitality. Conventional statistical analysis shows weak or not significant correlations between tree crown defoliation and climatic conditions or air pollution parameters, because of its high variability. The study aims to evaluate the most important factors among climatic, pollutants (Nox and NHy) and stand parameters affecting crown defoliation of the main European tree species (Fagus sylvatica, Picea abies, Quercus ilex, Pinus sylvestris and Quercus petraea) through application of a new and powerful statistical classifier, the random forests analysis (RFA). RFA highlighted that tree crown defoliation was mainly related to age in P. abies, to geographic location in F. sylvatica and to air pollution predictors in Q. ilex, while it was similarly linked to meteorological and air pollution predictors in P. sylvestris and Q. petraea. In this study, RFA has proven to be, for the first time, a useful tool to discern the most important predictors affecting tree crown defoliation, and consequently, it can be used for an appropriate forest management. © Springer International Publishing 2014.


cited By 7

Citation KeyVitale2014